Modeling “Presentness” of Electronic Health Record Data to Improve Patient State Estimation (2018)

by Jacob Fauber and Christian R. Shelton

Abstract: Medical data are not missing at random. The problem is more acute when the observations are over an extended period of time; any particular variable is observed at relatively few time points. We take missing values to be the norm, and treat “presentness” (the times of observations) as additional features to augment the values observed. A joint model over both avoids the missing-at-random (MAR) assumption. We use piecewise-constant conditional intensity models (PCIMs) to build a generative model of observation times and values. We demonstrate its effectiveness in reconstruction of monitor readings of patient vitals from sparse EHR data.

Download Information

Jacob Fauber and Christian R. Shelton (2018). "Modeling “Presentness” of Electronic Health Record Data to Improve Patient State Estimation." Proceedings of Machine Learning for Healthcare. pdf        

Bibtex citation

@inproceedings{FauShe18,
   author = "Jacob Fauber and Christian R. Shelton",
   title = "Modeling ``Presentness'' of Electronic Health Record Data to Improve Patient State Estimation",
   booktitle = "Proceedings of Machine Learning for Healthcare",
   booktitleabbr = "MLHC",
   year = 2018,
}

full list